The Most Reliable Hard Drive?

Hi Bob--I noticed that solid state hard drives don't seem to be mentioned. Are these more reliable?

Unless I missed it you did not comment on SSD Drives and their reliability rates. My understanding is they are initially more reliable but in the long term actually "wear out" even more than mechanical drives (although I understand "wear out" may be technically inaccurate. Any thoughts you might like to share in conjunction with the above article would be appreciated.

EDITOR'S NOTE: It appears that newer hard drives (from the last few years) are much more reliable than older ones. There's no longer any concern about them "wearing out."

See http://askbobrankin.com/ssd_drives_good_for_the_long_haul.html

Sorry Bob,
I had a nightmare with IBM Deskstar’s where not one, but two failed on me in quick succession, and lost four years of valuable multi-track music files because of them… and I wasn’t the only one who suffered from this Trabant of hard drives. Never again will I ever buy another Deskstar no matter what brand labeling it comes under. Once bitten a million times shy!!!

Hi Bob:

Interesting article, but I was a bit surprised that Western Digital was not on your or BackBlaze's recommended list. WD has a generally excellent reputation for reliability over many years, and is the brand that I usually purchase for that reason - plus my own positive experience with them.

The problem here seems to be that BackBlaze has not purchased very many WD drives (solely for cost reasons) - so the sample size for their reliability data is too small to give meaningful results.

Furthermore we see (in the BackBlaze graphs) that many of the Seagate 3 TB drives in BackBlaze's arrays have reached annual failure rates of over 40%! To me, this is unacceptable - despite a small price advantage.

Finally, BackBlaze is buying large quantities of drives at a time - but for consumers like us - a small difference in price is not a major factor in making a buying decision. I'm going to stick with Western Digital!!

I'll stick with Seagate 'Barracuda' because a hard disk named after a fast aggressive fish is all the assurance I need. No, really. I've tried buying other drives... The just don't have the Caché (drumroll) I'm here all week.

We have about 40 drives in service.
We use Hard Disk Sentinel as a monitor.
Our Seagate experience is horrible...many failing in less than a year.
Our Western Digital & Hitachi drives typically go 4 or 5 years before decaying to the point where they need replaced.
Have not bought a Seagate (Barracuda was once our favorite)in several years.

PS:: For those asking about SSDs, they are not Hard drives...it's an Apples & Orange thing...one is mechanical, the other is chemical..

I am surprised you didn't mention a well known Google's report about hard drives reliability that covered much more more brands and models, that in that article. The main conclusion was that there is no such thing as the best drive. Statistically any drive has the same reliability regardless of manufacturer or model (among those used by Google, of course). Also the SMART appeared to be pretty unreliable in case of failure prediction.

The report can be found there:

static.googleusercontent.com/media/research.google.com/en/us/archive/disk_failures.pdf

EDITOR'S NOTE: The Google study is from 2007. Hardly relevant 7 or 8 years later, as technology changes so rapidly. What are the chances that any of those drives are still in use? And I suspect they were not consumer-grade drives like BackBlaze uses.

Strange I didn't catch manufacturers list there - I remember they also mentioned them just after the research.

Google also claimed that they researched over 100,000 drives, but analysis of statistic data gave 830,000 drives that is much more, than "Backblaze had 41,213 disk drives in service".

Also I see you mentioned "HGST (formerly Hitachi)" that also was formerly IBM, well known by using glass disk plates instead of metal. As I remember, they work fine in stable environment, but even temperature fluctuations could led to surface damage. At least a dozen years ago I heard that cause from professionals in rescue drive's info. Didn't check it myself, so maybe it's one of urban legends. ;)

I saw a note about SSD. One of network administrators wrote his personal impression of that matter. His personal choice was Samsung Pro series drives, Intel and OCZ (as I remember). It was dated a couple years ago and he mentioned only drives that worked to that date a couple of years. Also he had some bad experience in info recovery - if you have a problem at SSD, it's done. In case of hard drive you has a chance to read info from surface with pretty special equipment that is impossible for SSDs since they are technically not drives, but memory arrays. So his rule of thumb is to have mandatory backup of those drives or use them for system, swap, pagefile or index files only.

Ouch, I forgot to note that there is no correlation among manufacturers, but there is a correlation among models - some of them are just bad. It could be even a faulty firmware version, or technological process mistake like it was with Fujitsu drives many years ago (something about soldering of control board, as I remember).

My personal opinion for what it is worth, I don't believe that newer is always better. I don' think that I need all the bells and whistles to be able to do what I want to do on my computer.
I do like the advances in technology, but I like to
be able to choose what I feel is a advantage and what is a sales pitch for something I don't need.
I still believe the best computer is between my own two ears.
From reading your articles, Your articles have confirmed that my opinions are not too far off.
I think anyone can read what is on the lines,but there is always as much information between the lines,if you can see it.
As always you say a lot between the lines. You do a excellent job of explaining things in a way us less knowledgeable can understand. A great job Bob.

Bob, are you serious about hard drive technology? The basic things of that technology are the same, as it was a dozen years ago, and even earlier - the same mechanical principle, the same magnetic surface, the same head block and linear and regular motors for head block and spindle respectfully. Of course, there are some updates, but basics are the same, as the problems and statistics. That is the main problem here - the technology is almost perfect for pretty long time, no room for serious breakthrough, just for improvements. Believe me, I have Master of Science in both Computer Science and Mechanical Engineering. Try to compare two disks of Google research date and today internally - you won't see the difference by a naked eye, new things are mostly in algorithms and interfaces. If you read the document you would catch, what kind of disks they used - both special and regular. They specified that in their document:

"The data in this study are collected from a large number of disk drives, deployed in several types of systems across all of Google’s services. More than one hundred thousand disk drives were used for all the results presented here. The disks are a combination of serial and
parallel ATA consumer-grade hard disk drives, ranging in speed from 5400 to 7200 rpm, and in size from 80 to 400 GB. All units in this study were put into production in or after 2001. The population contains several models from many of the largest disk drive manufacturers and from at least nine different models. The data used for this study were collected between December 2005 and August 2006" (the very beginning of Chapter 2.2)

Also you would be surprised by similarities in results - nothing new in this mush less fundamental research, sorry.

Old discoveries not necessary are obsolete - Newton laws of motion of classic physics are still actual, aren't they? Same with this old technology.

EDITOR'S NOTE: The Google study says "units put into production in or after 2001...ranging in size from 80 to 400 GB." We now have drives that hold 20 to 100 TIMES MORE DATA in the same (or smaller) physical spaces. That involves a lot of physics and new technology, Sure, they're all spinning magnetic platters, but the way they are designed and manufactured is quite different now than in 2001-2005. There are new techniques to improve areal density (cramming more data into smaller spaces), requiring new ways to overcome the resultant thermally induced magnetic instability. Other emerging technologies such as shingled magnetic recording, and heat-assisted magnetic recording promise even greater gains. I just don't see how you can compare 12-year-old apples to today's oranges.

Though interesting, there is a huge difference between how a typical home user operates a hard disk and how Backblaze uses them - I'm sure Backblaze runs their HDs 24/7 and seldom turns them off. The home user probably has the HD running a few hours a day, and, if not shut off, is spun down when sleeping or hypernating. The spin up / spin down cycles have an effect on life. (good or bad? experts opinions vary!) On the other hand, Backblaze probably works their drives harder reading and writing constantly.
I agree with kirill that drive *models* make a bigger difference; a previous 2014 Backblaze report really came down on Seagate 1.5 TB drives with a astounding 14% failure rate. The average home user would much more likely have one of these rather than a 4 TB drive.

oh ... and doing more checking. The Sep 2014 Backblaze update of HD reliability calls out the Seagate 1.5 TB (model ST31500341AS) for even worse reliability, a 24.9% annual failure rate. The best drive in this report was the HGST HDS5C3030ALA630 at only 0.7%.
The are wide differences within models from the same manufacturer.

I have only had one Hitachi drive to fail, and it was only two months old. That was very disappointing, however I have had five Seagate drives to fail that were less than six months old each. Sometimes I think it is just a matter of luck when it comes to failures. In my opinion, I don't think any hard drive should fail if it is less than one year old.

Well, but similarities in result tell quite opposite, since all innovations are about magnetic surface and that didn't improve main problem of faults. I see the results of all those innovations - the drives life is similarly short, like it was 12 years ago and even shorter, that it was 20 years ago - I remember perfectly worked over 15 years 210 Mb drive from Alps. According to your logic those drives employed better technology since they worked longer, do they? Sorry, Bob, but you are caught by good marketing, the problem with mechanical hard drives that they basically are the same and because of that have similar problems plus problems with super high density of info on them. And similarities in results just proved that one more time. I just see how easily modern drives die, taking with them into grave much more info, than it happen before with the same symptoms, like "click of death", stock spindle, fallen head unit and faulty sectors. Nothing new for me. I really don't see serious improvements that all those hi-tech bells and whistles do for reliability. Like it was way before, if you have a hardware problem, it mostly is a hard drive problem. Don't you surprise that there is no researches about reliability of motherboards, RAM, peripheral cards? That is the answer for all innovations in hard drive field - there is no breakthrough in reliability. I see the final result and don't like it. But we all live with it, surviving different ways - to see the problem or to fall for marketing tricks.

I'll stick to the 6 SAS drives in raid array to keep me happy. One goes bad I get notified and replace and it rebuilds array automaticly. No fuss no mess and don't have to worry about losing all of my valuable data. Yes, it's high end but if you don't want to lose it then you've got to be like the Boy Scouts and always be prepared!!!